AI Exploration: Puzzle Counter
The Origin of the Idea
My first real foray into AI-first design had its genesis in an offhand conversation my wife and I were having after she had returned from a charity shop with a new puzzle. I watched as she eagerly spread all of the pieces out on the table in preparation for a marathon of puzzling. She was meticulous in her setup, ensuring each piece was right-side up and in a single layer, and then spent the next half hour slowly working on the puzzle from the outside in. At that point, it started to become clear to her that the puzzle was wasn’t actually complete. There were clearly some gaps with no pieces to fill them.
Re-enactment of the exact moment Christina realized some of the pieces weren’t there.
She had spent about an hour preparing and then beginning the puzzle before having to deflatedly take apart the partially-complete puzzle and put it back into the box before relegating the once-coveted puzzle to the trash. She took it in stride, but this disappointment had the opposite effect as the zen-like act of methodical piece-by-piece creation she was seeking when buying the puzzle.
Seeing this and knowing that it wasn’t the first time, I had the very designer thought of “what if there was a way to avoid this and what would that look like.” Conveniently, I was also currently thinking about projects that would work well to support a foray into AI-first design. An app that had a limited purpose, such as counting puzzle pieces seemed like a great option since it was concrete with a constrained scope, had enough complexity that there could be meaningful differences and assumptions across the different tools, and solves a real user problem that that I could judge the systems’ output against.
My Approach
What I hope to do with this project is to use AI tools across all steps of an end-to-end design process from wireframe to prototype (with the stretch goal of building it in code). I am hoping that by the end of this process I will have used a variety of AI tools for ideation, wireframes, visual design, prototypes, PRDs, and code. The end goal will be to not only have a pretty cool app design (🤞), but also have a good sense of:
What tools are out there and what tasks they are best suited for
What is the overall maturity of these AI design tools
How AI might be integrated into existing design workflows vs. whether we would benefit from an overall rethinking of how we approach our projects
Start to piece together where less experienced designers fit into the process when things like wireframing can start to be done much more quickly by AI tools. In other words, how can I provide teachable moments when some of the work will be done by a black box
Although the discrete steps may look similar to what we’ve done in the past, AI promises to amplify our efforts from initial research and ideation through to prototyping and delivering. The question will be how to best couple AI’s speed and volume of output with the designers mastery of context, taste, and humanity. Source: The Design Council
Ultimately, this project will give me an opportunity to look deeply at some powerful and very exciting new tools that will surely become part of how we do things. I think it is very clear that, even if the steps of the design process continue on as normal, there is a new and powerful participant that will be at the table to drastically expand our ability to explore and ideate quickly and prototype and even build in ways we haven’t been able to do before. This will be especially true for smaller orgs and design teams who may not have had the headcount to involve multiple designers and technologists on every project.
¡Ay, ay, AI!
Over the next few weeks, I am going to use this blog to capture some of my thoughts and experiences diving a bit deeper into AI. Up to this point, I have been a pretty casual user of AI. I would use it to help draft emails, cover letters, and the like, or use tools like Perplexity to provide richer summaries of technical or historical topics.
I tried using it in my job, but my few forays into using it for work always left me feeling a bit underwhelmed. Even though I worked in a pretty well-understood industry (video streaming), when I would try to use it to work through use cases, or suggest feature improvements, the output always felt junior-level at best. The use cases were ones we had already come up with ourselves or were irrelevant to what we were working on, and the feature improvements were similarly limited to things that any member of our team would have come up with.
Given this, I have been a bit struck by the absolute mania that has taken over the discourse on LinkedIn and other design channels. Somewhere along the lines, we have gone from “AI could be a useful tools in our craft” to “let’s throw all of our chips in on AI practicing the craft.” I tend to be a fairly skeptical person, and suspect that the stridency and absolutism I am seeing is a product of professional anxiety and insecurity, moreso than any indication that AI is now capable of creating design of great elegance or sophistication. Which is not to say that that anxiety and insecurity isn’t justified. Many of these tools implicitly, or sometimes even explicitly, suggest that our role as designers can now be done by a Product Manager+AI, and I have been around long enough to know that justifying the work we do has always been a major part of the job and we haven’t ever had competition quite like this.
This was the first image ChatGPT spit out, and is perhaps the perfect AI generated image to capture the themes of my exploration. No notes.
So, with that in mind, I am going to dive into designing with AI to explore what is out there and investigate how these new and powerful tools might serve the craft I love and help me to become a better designer. My hope is that by exploring what is out there, using it in my design practice, and putting together some workflows, I will be able to better identify where the use of AI will improve the craft of design, and begin to grapple with some of my concerns about how, without major changes to how we operate design teams, we will train young designers to think through design problems rather than offloading that initial work to AI.